A regression approach for Zircaloy-2 in-reactor creep constitutive equations

Typical data analysis procedures used in developing current phenomenological in-reactor creep equation for Zircaloy cladding materials are examined. It is found that the data normalization assumptions and the curve fitting techniques generally adopted in these procedures can make the prediction of creep strain rate highly questionable. Multiple regression analysis performed on a set of carefully selected Zircaloy-2 in-reactor creep data on the basis of comparable as-fabricated metallurgical conditions indicates that models of the form creep strain rate = A(sigma to the n power)(phi to the m power) exp(-Q/RT) can give significant conditions for the data. Both regression statistics and residual plots have provided strong evidences for the significance of the regression equations.